I'm not sure what points you'll end up making (wrt point #2), but I just want to state that this approach makes sense to me. It's often productive to think through simple cases, or cases where our data are better, and then try to think explicitly about whether the lessons learned there should carry over to other cases that are in some ways more difficult to analyze.
In philosophy, a common saying is "It's best not to try to do philosophy all at once." Very much in the spirit of this comment and Jonah's point #3.
My recent posts Robustness of Cost-Effectiveness and Effective Philanthropy and Earning to Give vs. Altruistic Career Choice Revisited concern optimal philanthropy, and I’ll be writing more posts about optimal philanthropy in the near future.
My use of examples from prosaic domains such as global health has given rise to some confusion, because some members of the Less Wrong community believe that existential risk reduction is by far the best target for optimal philanthropy, and also believe that effective philanthropy in the context of global health is very disanalogous to effective philanthropy in the context of x-risk reduction. For example, Eliezer wrote
I believe that studying the issues surrounding philanthropic opportunities in areas such as global health is in fact helpful for better understanding how to assess x-risk reduction opportunities. My reasons for thinking this don’t fit into a few sentences, and fully understanding them requires understanding some of my thoughts about more prosaic domains. So I’ll respond to Eliezer’s comment at a later date.
For now, I’ll just remark:
Note: I formerly worked as a research analyst at GiveWell. All views expressed are my own.